File size: 10,134 Bytes
e72d7b7 b73125b e72d7b7 da76781 e72d7b7 a2f95d7 e72d7b7 da76781 e72d7b7 a2f95d7 e72d7b7 bbd06c4 5ec32be bbd06c4 5ec32be bbd06c4 e72d7b7 3e1a621 a2f95d7 e72d7b7 5ec32be e72d7b7 a2f95d7 e72d7b7 a2f95d7 e72d7b7 a2f95d7 a712c99 a2f95d7 bbd06c4 a2f95d7 e72d7b7 a2f95d7 e72d7b7 bbd06c4 e72d7b7 8addf64 e72d7b7 a2f95d7 e72d7b7 a712c99 5ec32be a2f95d7 5ec32be b73125b bbd06c4 5ec32be bbd06c4 37686b3 b73125b 37686b3 0fcf8a1 a6c1bac 0fcf8a1 37686b3 b73125b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 |
from huggingface_hub import InferenceClient, HfApi, upload_file
import datetime
import gradio as gr
import random
import prompts
import json
import uuid
import os
token=os.environ.get("HF_TOKEN")
username="omnibus"
dataset_name="tmp"
api=HfApi(token="")
client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
history = []
hist_out= []
summary =[]
main_point=[]
summary.append("")
main_point.append("")
def format_prompt(message, history):
prompt = "<s>"
for user_prompt, bot_response in history:
prompt += f"[INST] {user_prompt} [/INST]"
prompt += f" {bot_response}</s> "
prompt += f"[INST] {message} [/INST]"
return prompt
agents =[
"COMMENTER",
"BLOG_POSTER",
"REPLY_TO_COMMENTER",
"COMPRESS_HISTORY_PROMPT"
]
temperature=0.9
max_new_tokens=256
max_new_tokens2=10480
top_p=0.95
repetition_penalty=1.0,
def compress_history(formatted_prompt):
seed = random.randint(1,1111111111111111)
agent=prompts.COMPRESS_HISTORY_PROMPT.format(history=summary[0],focus=main_point[0])
system_prompt=agent
temperature = 0.9
if temperature < 1e-2:
temperature = 1e-2
generate_kwargs = dict(
temperature=temperature,
max_new_tokens=30480,
top_p=0.95,
repetition_penalty=1.0,
do_sample=True,
seed=seed,
)
#history.append((prompt,""))
#formatted_prompt = format_prompt(f"{system_prompt}, {prompt}", history)
formatted_prompt = formatted_prompt
stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
output = ""
for response in stream:
output += response.token.text
#history.append((output,history))
print(output)
print(main_point[0])
return output
def question_generate(prompt, history, agent_name=agents[0], sys_prompt="", temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0,):
#def question_generate(prompt, history):
seed = random.randint(1,1111111111111111)
agent=prompts.COMMENTER.format(focus=main_point[0])
system_prompt=agent
temperature = float(temperature)
if temperature < 1e-2:
temperature = 1e-2
top_p = float(top_p)
generate_kwargs = dict(
temperature=temperature,
max_new_tokens=max_new_tokens,
top_p=top_p,
repetition_penalty=repetition_penalty,
do_sample=True,
seed=seed,
)
#history.append((prompt,""))
formatted_prompt = format_prompt(f"{system_prompt}, {prompt}", history)
stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
output = ""
for response in stream:
output += response.token.text
#history.append((output,history))
return output
def blog_poster_reply(prompt, history, agent_name=agents[0], sys_prompt="", temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0,):
#def question_generate(prompt, history):
seed = random.randint(1,1111111111111111)
agent=prompts.REPLY_TO_COMMENTER.format(focus=main_point[0])
system_prompt=agent
temperature = float(temperature)
if temperature < 1e-2:
temperature = 1e-2
top_p = float(top_p)
generate_kwargs = dict(
temperature=temperature,
max_new_tokens=max_new_tokens,
top_p=top_p,
repetition_penalty=repetition_penalty,
do_sample=True,
seed=seed,
)
#history.append((prompt,""))
formatted_prompt = format_prompt(f"{system_prompt}, {prompt}", history)
stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
output = ""
for response in stream:
output += response.token.text
#history.append((output,history))
return output
def create_valid_filename(invalid_filename: str) -> str:
"""Converts invalid characters in a string to be suitable for a filename."""
invalid_filename.replace(" ","-")
valid_chars = '-'.join(invalid_filename.split())
allowed_chars = ('a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm',
'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z',
'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M',
'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z',
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '_', '-')
return ''.join(char for char in valid_chars if char in allowed_chars)
def load_html(inp):
ht=""
for ea in inp:
outp,prom=ea
ht+=f"""<div>
<pre>{outp}</pre>
<pre>{prom}</pre>
</div>"""
with open('index.html','r') as h:
html=h.read()
html = html.replace("$body",f"{ht}")
h.close()
return html
def generate(prompt, history, agent_name=agents[0], sys_prompt="", temperature=0.9, max_new_tokens=1048, top_p=0.95, repetition_penalty=1.0,):
html_out=""
#main_point[0]=prompt
#print(datetime.datetime.now())
uid=uuid.uuid4()
current_time = str(datetime.datetime.now())
current_time=current_time.replace(":","-")
current_time=current_time.replace(".","-")
print (current_time)
agent=prompts.BLOG_POSTER
system_prompt=agent
temperature = float(temperature)
if temperature < 1e-2:
temperature = 1e-2
top_p = float(top_p)
hist_out=[]
sum_out=[]
json_hist={}
json_obj={}
full_conv=[]
filename=create_valid_filename(f'{prompt}---{current_time}')
post_cnt=1
while True:
seed = random.randint(1,1111111111111111)
if post_cnt==1:
generate_kwargs = dict(
temperature=temperature,
max_new_tokens=max_new_tokens2,
top_p=top_p,
repetition_penalty=repetition_penalty,
do_sample=True,
seed=seed,
)
if prompt.startswith(' \"'):
prompt=prompt.strip(' \"')
formatted_prompt = format_prompt(f"{system_prompt}, {prompt}", history)
post_cnt+=1
else:
system_prompt=prompts.REPLY_TO_COMMENTER.format(focus=main_point[0])
generate_kwargs = dict(
temperature=temperature,
max_new_tokens=max_new_tokens2,
top_p=top_p,
repetition_penalty=repetition_penalty,
do_sample=True,
seed=seed,
)
if prompt.startswith(' \"'):
prompt=prompt.strip(' \"')
formatted_prompt = format_prompt(f"{system_prompt}, {prompt}", history)
if len(formatted_prompt) < (50000):
print(len(formatted_prompt))
stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
output = ""
#if history:
# yield history
for response in stream:
output += response.token.text
yield '', [(prompt,output)],summary[0],json_obj, json_hist,html_out
out_json = {"prompt":prompt,"output":output}
prompt = question_generate(output, history)
#output += prompt
history.append((prompt,output))
print ( f'Prompt:: {len(prompt)}')
#print ( f'output:: {output}')
print ( f'history:: {len(formatted_prompt)}')
hist_out.append(out_json)
#try:
# for ea in
with open(f'{uid}.json', 'w') as f:
json_hist=json.dumps(hist_out, indent=4)
f.write(json_hist)
f.close()
upload_file(
path_or_fileobj =f"{uid}.json",
path_in_repo = f"book1/{filename}.json",
repo_id =f"{username}/{dataset_name}",
repo_type = "dataset",
token=token,
)
else:
formatted_prompt = format_prompt(f"{prompts.COMPRESS_HISTORY_PROMPT.format(history=summary[0],focus=main_point[0])}, {summary[0]}", history)
#current_time = str(datetime.datetime.now().timestamp()).split(".",1)[0]
#filename=f'{filename}-{current_time}'
history = []
output = compress_history(formatted_prompt)
summary[0]=output
sum_json = {"summary":summary[0]}
sum_out.append(sum_json)
with open(f'{uid}-sum.json', 'w') as f:
json_obj=json.dumps(sum_out, indent=4)
f.write(json_obj)
f.close()
upload_file(
path_or_fileobj =f"{uid}-sum.json",
path_in_repo = f"book1/{filename}-summary.json",
repo_id =f"{username}/{dataset_name}",
repo_type = "dataset",
token=token,
)
prompt = question_generate(output, history)
main_point[0]=prompt
full_conv.append((output,prompt))
html_out=load_html(full_conv)
yield prompt, history, summary[0],json_obj,json_hist,html_out
return prompt, history, summary[0],json_obj,json_hist,html_out
with gr.Blocks() as app:
html = gr.HTML()
chatbot=gr.Chatbot()
msg = gr.Textbox()
with gr.Row():
submit_b = gr.Button()
stop_b = gr.Button("Stop")
clear = gr.ClearButton([msg, chatbot])
sumbox=gr.Textbox("Summary", max_lines=100)
with gr.Column():
sum_out_box=gr.JSON(label="Summaries")
hist_out_box=gr.JSON(label="History")
sub_b = submit_b.click(generate, [msg,chatbot],[msg,chatbot,sumbox,sum_out_box,hist_out_box,html])
sub_e = msg.submit(generate, [msg, chatbot], [msg, chatbot,sumbox,sum_out_box,hist_out_box,html])
stop_b.click(None,None,None, cancels=[sub_b,sub_e])
app.load(load_html,None,html)
app.launch() |